Schedule 37th Chaos Communication Congress

I'm a researcher in the field of machine learning hardware and compression of deep learning models for inference on microcontrollers and other small devices. I fiddle with embedded systems (bonus points if they contain LEDs) in my spare time.

I am currently pursuing a Ph.D. on the topic of deploying deep learning models on devices that are severely constrained in terms of energy, compute and memory (think: embedded microcontrollers that run on batteries).
I work in the areas of:

• Pruning: trying to reduce a deep learning model to its most useful parameters and dropping the rest in order to shrink it


• Approximate Computing: Using hardware that does not guarantee mathematically accurate results in an effort to reduce their arithmetic complexity

I'm mostly interested in modelling the effects of constrained hardware during the training of deep learning networks so that models can learn themselves how to deal with modest hardware.
Also always happy to have a chat about embedded systems in general, PCB design, and all things electronics. Card-carrying member of the Rust evangelism strike force.

Links

• Personal website
• Github